Best possible solution:

Review and land exactly one proof-positive parser-boundary fix, preferably #99399, then close the linked issue and remaining duplicate PRs.

Do we have a high-confidence way to reproduce the issue?

Yes. Current main clearly parses "1.5" through the single-token Number(part) branch, and fallback extraction filters non-integer page numbers later.

Is this the best way to solve the issue?

Yes for the fix layer, but this branch is not the best landing vehicle. The parser-boundary integer check is correct, while #99399 is the stronger canonical PR to review and land.

Security review:

Security review cleared: The diff only tightens PDF parser validation and adds tests; it does not touch dependencies, workflows, secrets, package metadata, or supply-chain execution paths.

AGENTS.md: found and applied where relevant.

What I checked:

• Current parser accepts fractional single-page tokens: On current main, parsePageRange converts single tokens with Number(part) and only rejects non-finite or values below 1, so "1.5" can still enter the page-number contract. (src/agents/tools/pdf-tool.helpers.ts:68, f296dc4e68e4)
• PDF tool uses the parser as the shared boundary: createPdfTool computes pageNumbers once from parsePageRange and passes it to both extraction and model routing, so the parser is the right fix layer. (src/agents/tools/pdf-tool.ts:498, f296dc4e68e4)
• Fallback extraction filters non-integer pages: The document-extract plugin filters requested pages with Number.isInteger, so a fractional parser output can be dropped downstream instead of reported as invalid input. (extensions/document-extract/document-extractor.ts:68, f296dc4e68e4)
• Dependency contract also expects integer pages: The pinned [email protected] package validates loaded page numbers with Number.isInteger and documents one-based page APIs, matching integer-only OpenClaw parsing.
• Canonical open PR is viable: fix(pdf): reject fractional page selections #99399 is open, cleanly mergeable, has proof: sufficient, and covers both direct fractional tokens and mixed comma-separated fractional input before fallback extraction. (451ef0407653)
• Current branch duplicates the same parser fix: This PR changes the same helper and adds narrower parser/native-path tests for the same linked issue, without a unique implementation or validation advantage over the canonical candidate. (src/agents/tools/pdf-tool.helpers.ts:66, c18f5e2d6d8d)

Likely related people:

• tyler6204: Remote history shows the original merged PDF analysis tool PR added parsePageRange, the public pages argument, and the main PDF tool implementation. (role: introduced behavior; confidence: high; commits: d0ac1b019517; files: src/agents/tools/pdf-tool.helpers.ts, src/agents/tools/pdf-tool.ts, src/agents/tools/pdf-tool.test.ts)
• zhangguiping-xydt: Authored the recently merged neighboring fix for empty parsed PDF page selections in the same parser helper and tests. (role: recent adjacent repair author; confidence: high; commits: a5165d6a8486, aadd57c31cc8; files: src/agents/tools/pdf-tool.helpers.ts, src/agents/tools/pdf-tool.helpers.test.ts, src/agents/tools/pdf-tool.test.ts)
• steipete: Recent remote history includes multiple PDF tool/helper and test refactors around the affected parser and tool surface. (role: recent area contributor; confidence: medium; commits: 7240830ca42f, 998cc02af40c, 1ce11fbf423b; files: src/agents/tools/pdf-tool.helpers.ts, src/agents/tools/pdf-tool.ts, src/agents/tools/pdf-tool.helpers.test.ts)
• vincentkoc: Merged the adjacent empty-page parser fix and appears in recent PDF/document-extract history around this surface. (role: recent merger and adjacent contributor; confidence: medium; commits: aadd57c31cc8, 3881cb34260a; files: src/agents/tools/pdf-tool.helpers.ts, src/agents/tools/pdf-tool.ts, extensions/document-extract/document-extractor.ts)

Codex review notes: model internal, reasoning high; reviewed against f296dc4e68e4.